AI safety is designed in the West – failing users in the rest of the world
AI safety is primarily designed in the West and fails to protect users globally, according to an investigation. Cultural and linguistic oversights are creating significant security gaps.
What happened?
According to an investigation by Rest of World, current AI safety systems are failing users outside the Western world. Safety protocols and guardrails are primarily designed and evaluated based on Western values and linguistic patterns. This causes AI models to frequently miss harmful content in smaller languages, while simultaneously incorrectly censoring benign local communication.
Key facts
| Källa till granskning | Rest of World |
|---|---|
| Fokusområde | Global AI-säkerhet och språkmodeller |
Why it matters
The problem stems from the fact that training data and red-teaming (security testing) are primarily conducted in English and based on North American or Western European norms. As technology companies expand globally, cultural and linguistic contextualisation is often missing, creating security gaps in moderation and a lack of reliability for local users.
Who is affected?
The situation affects millions of users in the Global South and non-English-speaking countries who use AI tools from leading tech companies. It also affects developers and researchers working on adapting AI models for local markets and languages.
Impact on the EU
In Europe, AI is increasingly regulated through the AI Act and GDPR, which impose high requirements for transparency and risk analysis. However, the issue of Western-centric safety models also affects smaller language groups and minorities within the EU, as training and evaluation are often dominated by English-language standards.
What else you should know
The report emphasises that achieving globally functional AI safety requires leading AI companies to invest in locally grounded testing methods and include more multifaceted data from various linguistic regions in their evaluation processes.
Quick answers about this story
Vad har hänt?
När hände det?
Varför spelar det roll?
Vilka berörs av bristerna?
The link opens in a new window and leads to the publisher's own site.
Källan är en aggregator eller syndikering — vi rekommenderar att verifiera hos primärutgivaren.
AI-verktyg i artikeln
Topics
Get similar news straight to your inbox
The reader's room
Send in a question or an addition. The newsroom reads everything before it's published and replies when relevant. No AI-generated text – just people.
Sign in to submit a comment or question.
Read the article through your role
- Decide whether this affects strategy over 6–12 months or is just noise.
- Discuss with leadership: do we own the right question or does ownership need to move?
- Ask: what risk are we taking by NOT acting on this this quarter?
Generated angle — not editorial analysis of "AI safety is designed in the West – failing users in the res"